This project investigates late-learned second language (L2). Adult subjects will learn an artificial language to low and then high proficiency, under either explicit or implicit training conditions. Behavioral and neuroimaging measures (ERP, fMRI) employed at both the low and high proficiency levels within each subject will elucidate neurocognitive correlates of L2. The predictions of competing neurocognitive models and of competing applied linguistic models of L2 acquisition and processing will be examined. The specific hypotheses are derived from the declarative/procedural neurocognitive model. This model posits that in L1, the mental lexicon of memorized words depends on the temporal-lobe-based declarative memory system, whereas the mental grammar, which underlies the composition of complex structures, relies on the frontal/basal-ganglia based procedural memory system. For L2, the model predicts that, at least at lower levels of proficiency, the processing of complex forms relies largely on declarative memory. It is expected, however, that practice should lead to an increased reliance on the procedural system for the learning and processing of complex structures, as in L1. Recent evidence supports these predictions. The results of the project are expected to confirm these neurocognitive predictions, and to distinguish between applied linguistics models of L2 regarding the role of explicit vs. implicit training. Results of the project may have important implications for L2 learning and for the recovery of bilingual language-impaired populations. [unreadable] [unreadable]